Marketers are very creative people. There is an art to designing messaging, copy, ads, websites and a long list of other assets — and pulling them into a cohesive strategy.
And that strategy must achieve more than aesthetic appeal. Something might look nice or read well, but it’s a waste of time if it does not help achieve business results.
This is why marketers need to add science to their art. Through experimentation, marketers uncover the best designs and strategies to improve response rates.
For example, marketers using an inbound marketing strategy must test their landing pages to improve the conversion rates of incoming traffic (check out MarketingSherpa’s first InBound Marketing newsletter, launched yesterday). Ignoring experiments will waste traffic and resources.
Choose the Best Test
Back in elementary science, we all learned about the Scientific Method — the process used to answer questions with experiments and observations. You remember: Ask a question, form a hypothesis, test the hypothesis, etc.
One thing they didn’t teach us in school is that marketers need to prioritize their experiments to efficiently use their resources and achieve the best possible results.
Let me explain:
In science, very precise experiments measure the results of changes to a single variable while holding all other factors constant. This is a great way to discover that single variable’s role.
In marketing — and in landing page optimization specifically — a single variable can be a very specific detail, such as a button’s shape. Take a look at these two buttons:
An A/B landing page test comparing these buttons would actually change more than six variables. A button is a group of variables, or a variable cluster, which includes the button’s:
Bigger Tests for Better Results
Variable clusters are treated as single variables for the purpose of experiments. Some marketers believe their tests change a single variable when in fact they’re really clusters.
But this is OK!
“Testing multiple variables is more likely to increase the size of the results variation,” said Bob Kemper, Senior Director of Sciences, MECLABS, during a recent MarketingExperiments Web clinic. “Testing individual variables or smaller clusters is more likely to increase the precision of the test.”
Most marketers do not have the resources to isolate and test every variable on a landing page — there are far too many. Instead, they test larger changes, which are more likely to substantially alter their results — because time and money are a factor.
“Testing larger variable clusters… requires fewer treatments which in turn reduces the test duration and can substantially boost the return on your investment in time and energy,” Kemper said.
So while marketers should use science and experiments to improve performance — prioritization is the key. The goal is not to identify every perfect detail. The goal is to efficiently increase performance. Then, if you want, you can test isolated variables to learn more about your audience. And of course you will do it with an artistic touch.